RESEARCH ARTICLE | AUGUST 08 2023

Experimental and numerical studies of piston-mode

resonance in a three-dimensional circular moonpool 
Special Collection: Recent Advances in Marine Hydrodynamics

Jian Han (韩健) ; Xinshu Zhang (张新曙)  ; Haiyang Huang (黄海洋)

Physics of Fluids 35, 082106 (2023)

https://doi.org/10.1063/5.0160461


View Export
Online Citation

Articles You May Be Interested In

A numerical study of the influence of the wave reflection on the hydrodynamic characteristics of a cylinder
with moonpool
Physics of Fluids (January 2024)

On the nonlinear moonpool responses in a drillship under regular heading waves

Physics of Fluids (March 2024)

Effects of multi motion responses and incident-wave height on the gap resonances in a moonpool

27 December 2024 14:27:57

Physics of Fluids (January 2024)
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

Experimental and numerical

studies of piston-mode resonance
in a three-dimensional circular moonpool
Cite as: Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461
Submitted: 1 June 2023 . Accepted: 19 July 2023 .
Published Online: 8 August 2023

Jian Han (韩健),1,2,3 Xinshu Zhang (张新曙),1,2,a) and Haiyang Huang (黄海洋)1,2

AFFILIATIONS
1
State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
2
Collaborative Innovation Center for Advanced Ship and Deep-Sea Exploration (CISSE), Shanghai Jiao Tong University,
Shanghai 200240, China
3
Department of Ocean Technology Policy and Environment, Graduate School of Frontier Science, The University of Tokyo, 5-1-5,
Kashiwano-ha, Kashiwa, Chiba 277-8563, Japan

Note: This paper is part of the special topic, Recent Advances in Marine Hydrodynamics.
a)
Author to whom correspondence should be addressed: [email protected]

27 December 2024 14:27:57

ABSTRACT
We develop a modified potential flow model (MPFM) for a circular moonpool by quantifying the viscous damping. The flow-separation-
induced damping coefficients of MPFM do not need to be tuned by the users as they are derived from the conduit problem studied by
Idelchik, where the variation of such coefficients with geometry configurations is given based on experiments. To verify the MPFM, we
conducted laboratory experiments in heave radiation (forced harmonic heave motion) and wave diffraction (fixed in regular wave) problems
with a broad range of frequencies and compared the results with those of MPFM and a boundary element method without viscosity
modeling (BEM). The MPFM achieves fairly good agreement with the experimental results, including those around the resonance frequency
where BEM over-predicts. A simple time-domain model is developed to analyze the sources of the different harmonics in the free-surface ele-
vation observed in the experiment.
Published under an exclusive license by AIP Publishing. https://doi.org/10.1063/5.0160461

I. INTRODUCTION et al.7 and Zhang et al.8 developed new models to study the moonpool
The moonpool studied in the present paper can be the one inside resonance in finite water depth for three-dimensional and two-
a turret-disconnectable floating production storage and offloading dimensional cases, respectively. Natural frequencies of moonpools
(FPSO) (see Molin et al.1) or can model an opening in a floating pro- with more sophisticated shapes can also be investigated in a similar
duction unit, such as a Spar platform with an open centerwell,2 a mono- approach.9,10
column platform, a floating foundation for an offshore wind turbine,3 or Time-domain and frequency-domain potential flow models can
the chambers of dual oscillating water column devices.4 Both the natural study the shape of the free surface and hydrodynamic characteristics
frequencies and the resonant fluid motion inside the moonpool can be of the structure and thus can identify the resonance frequencies.
critical to the engineering practice. These subjects are important to the Faltinsen et al.11 investigated the two-dimensional piston-like wave
design and optimization of a floating structure with a moonpool in motion inside a moonpool based on a domain-decomposition scheme
order to avoid large resonant wave response, which can lead to negative and the Galerkin method and compared the numerical results with the
impacts to offshore operations and equipment on board.5 data from model tests. Mavrakos12,13 computed the wave load and the
The natural frequencies can be analyzed based on simplified added mass and damping coefficients of a truncated vertical cylinder
models, where explicit formulas for such frequencies may be devel- with a moonpool, respectively. Feng and Bai14 investigated the nonlin-
oped. Molin6 investigated natural frequencies and corresponding ear waves in the gap formed by two side-by-side arranged vehicles.
modal shapes of the free surface in two-dimensional and three- Song et al.15 investigated the transient responses and higher order har-
dimensional rectangular moonpools in infinite-depth water. Molin monics in the two gaps formed by closely placed bodies.

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-1

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

Recently, the recess has been adopted in the design of a moon- inlet of a two-dimensional moonpool, which contributes the majority
pool of a drill ship as it can provide space for assembling and trans- of viscous damping.22,23,25 The results agree well with the experiments.
porting the equipment, but it can significantly affect the Tan et al.29 adopted a similar approach and investigated the wave
hydrodynamic behaviors.16 Molin17 and Newman18 studied the prob- motions in a two-dimensional moonpool with rounded corners at the
lem of a moonpool with a recess, assuming that the flow inside the inlet. Many empirical formulas and corresponding coefficients for
moonpool is two-dimensional. Xu et al.19 studied the shift of the reso- pressure discharge over screens with various shapes are listed in
nance frequency of a moonpool with a recess in a free-floating barge, Blevins.30
which was investigated by Fredriksen et al.20 for a two-dimensional It should be noted that the above-mentioned studies are all
moonpool without recess. focused on two-dimensional cases. The damping effect due to flow sep-
Though the potential flow models can provide accurate reso- aration from the inlet of the three-dimensional moonpool (i.e., circular
nance frequencies, they usually over-predict the amplitude of free- or rectangular moonpool) is still an open question but is very impor-
surface elevation around there. Jiang et al.21 found that neglecting the tant for practical application in the offshore industry, which motivates
viscous damping may also lead to an underestimate of the wave ampli- the present study. Here, we aim to derive a robust and efficient model
tude outside the moonpool. Such a problem is overcome by applying to simulate the piston-like responses in a three-dimensional circular
computational fluid dynamics (CFD) codes with viscous solvers. A moonpool based on understanding and quantifying the damping
number of studies have been performed using viscous flow solvers effects due to the flow separation around the moonpool inlet.
(see, e.g., Kristiansen and Faltinsen,22,23 Heo et al.24) where the numer- The present paper is organized as follows. First of all, we compare
ical results show satisfactory agreement with the experiments and pro- the flow through a three-dimensional moonpool and a conduit
vide details of the flow fields. mounted in the wall, of which the flow-separation-induced pressure
However, the CFD computations are generally time-consuming; discharge was studied by Idelchik.31 Based on the similarity of flow
thus, are not suitable for parametric studies that are usually required fields, we propose a new modified potential flow model (named
in the design process for industry. The over-prediction of potential MPFM) by quantifying the flow-separation-induced pressure discharge
flow models is expected to alleviate if modifications reflecting the vis- and the friction force. Next, the experimental setup is described and
cous damping mechanism are implemented. As studied by Molin uncertainty analyses are presented in the corresponding Appendix A.
et al.,25 Molin and Remy,26 and Chen et al.,27 the conventional poten- Then, we compare the experimental and numerical results and discuss
tial flow model can be modified to account for viscous damping by the individual effects due to flow-separation-induced damping and fric-

27 December 2024 14:27:57

adding dissipation terms in the free-surface boundary condition. tion damping. Different harmonics of the free-surface elevation are
A major obstacle to this approach is that the dissipation coeffi- observed in the experiment and investigated by developing a simplified
cient itself is a priori that varies with the main dimensions of the time-domain numerical model based on the frozen-mode approxima-
moonpool and the floating structure. As a result, it may have to be tion. Finally, we summarize the results and give concluding remarks.
measured again when the geometric configuration is varied during the
II. MATHEMATICAL FORMULATION
design. Thus, obtaining generic dissipation coefficients or their depen-
dence on the geometric configuration so that users do not have to tune A. Statement of the problem
it frequently by dedicated experiments or CFD simulations is of crucial We consider the steady-state wave motions in the three-
importance. Based on the similarity of the flow fields, Faltinsen and dimensional moonpool of a circular cylinder. The sketch for the struc-
Timokha28 introduced an empirical formula of the pressure discharge ture foundation and the coordinate system is shown in Fig. 1. Let the
over a slatted screen to quantify the effects of flow separation at the Oxy plane be the calm-water surface, ðr; hÞ be the polar coordinates in

FIG. 1. Geometric sketch of the problem

and definition of the coordinate system.
Based on domain decomposition, three
subdomains are defined and two common
boundaries are denoted.

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-2

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

the horizontal plane, the z-axis coincide with the symmetry axis of the
moonpool and point upward, with the origin O of the Oxyz Cartesian
coordinate system and the polar coordinate system being located at
the center of the moonpool. The internal and external radii of the cyl-
inder are a and b, respectively, with ðb  aÞ being the wall thickness.
The draft of the vertical cylinder is d. The water depth is h. The corner
of the inlet of moonpool is sharp. Thus, flow separation is induced
there when the fluid flows into and out of the moonpool, which is one
of the sources of viscous damping studied in the present paper.
The wave motions in the moonpool can be induced when the cyl-
inder undergoes a forced heave motion in originally calm water (heave
radiation problem) or is fixed in the incident wave (wave diffraction
problem). The heave motion of the structure foundation and the free-
surface elevations of the incident wave are defined in Eqs. (B1) and
(C1) in Appendixes B and C, where n3a and Ai are the amplitudes of
the heave motion and incident wave, respectively, and r is the angular
frequency of the heave motion or incident wave. Here, we define the
non-dimensional amplitude of the forced heave motion and incident
wave by e ¼ n3a =a and e ¼ Ai =a, respectively.
The potential flow theory is applied based on the assumption
that the fluid is ideal and the flow irrotational. The linearized bound-
ary value problem for steady-state linear waves is formulated by FIG. 2. The comparison of the flow entering or exiting (a) a conduit mounted in the
wall and (b) a circular moonpool in a cylinder. The solid lines denote the identity of
the streamlines in the flow field, while the dotted lines denote the difference. The
DU ¼ 0 in Q0 ; (1)
shaded circle illustrates the area where flow separation is induced and disturbs the
@U flow. The dashed lines denote the flow that does not exist in the moonpool
¼0 on SB þ S; (2) (because the conduit is filled with fluid while there is a free surface in the moon-

27 December 2024 14:27:57

@n pool). The arrows show the directions of the streamlines either entering or exiting
@U the moonpool, not the flow direction at the same time.
¼ d3q n3a r sin ðrt Þ on SD ; (3)
@n
@U @g where f is the pressure loss coefficient that varies with the geometric
¼ on SF ; (4)
@z @t configurations of the conduit. The variation of f with geometric con-
@U figurations is shown in Diagram 3–1 in Idelchik31 (also, see Table II in
þ gg ¼ 0 on SF ; (5) the present paper). The other parameters and the geometric configura-
@t
  tions are shown in Fig. 2(a) and explained in Table I. Given any geo-
pffiffi @    
lim r  im0 U  1  d3q U0 ¼ 0 on SF ; (6) metric configurations of the conduit, we can find the pressure loss
r!1 @r
coefficient f based on Table II.
where Q0 is the fluid domain, SF is the calm-water surface on z ¼ 0, S Following Faltinsen and Timokha,28 if the flow fields of con-
refers to the mean wetted side walls of the cylinder, SB is the seabed, SD duit and moonpool are similar, it is reasonable to derive the results
is the bottom of the body, n is the outward unit normal to the fluid of Idelchik31 to model the pressure discharge in the moonpool
boundary, and d3q is the Kronecker delta, d3q ¼ 1 when q ¼ 3, i.e., the problem. We compare the flow fields of the flow entering the inlet
cylinder undergoes a forced heave motion; if the cylinder is fixed in of conduit and moonpool, as shown in Figs. 2(a) and 2(b), respec-
the incident wave, d3q ¼ 0. g represents the free-surface elevation. tively. We simplify the free surface outside the cylinder with the
Equation (6) is the radiation condition in the far field for both radia- moonpool as a wall, due to the low wave amplitude there compared
tion and diffraction problems, where m0 is the wavenumber of outgo- to that inside the moonpool [confirmed by the numerical simula-
2
ing waves defined by the dispersion relation m0 tanhðm0 hÞ ¼ rg , g is tions of MPFM and boundary element method (BEM)]. In Fig. 2,
the gravitational acceleration, and U0 is the velocity potential of the the shaded circle roughly marks the area where flow separation is
incident wave, whose spatial potential is defined in Eq. (C8). induced and disturbs the flow; the solid lines and dotted lines
denote the identity and difference of the streamlines in the flow
B. Numerical models of viscous damping field, respectively. The flows approaching the conduit and the
moonpool are almost the same and so are the flows inside the con-
1. Flow-separation-induced damping
duit and the moonpool. The flow fields near the inlets of the con-
Idelchik31 investigated the pressure loss of flow entering the con- duit and the moonpool are not the same. However, we can expect a
duit mounted in a wall. The sketch of the conduit problem is shown in high similarity in the flow fields where flow separation is generated
Fig. 2(a). The pressure loss is described by and disturbs the flow (denoted by shaded circles in Fig. 2) based on
the near identity in the flow fields both upstream and downstream.
f Based on the similarity, we describe the pressure discharge of the
DH ¼ cw2 ; (7)
2g 0 moonpool by

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-3

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

TABLE I. The explanation of the parameters and geometry configurations in Diagram 3-1 in Idelchik31 and the analogy between the conduit and moonpool problems.

Parameter in Diagram 3-1 Analogy in moonpool problem

B, the distance between the entrance of the conduit and the wall d, the draft of the moonpool
d, the thickness of the conduit ðb  aÞ
Dh, the hydraulic diameter 2a
w0, the speed of fluid in the conduit V, the relative speed of fluid in the moonpool and the cylinder
f, the pressure loss coefficient K, the pressure discharge coefficient
c, specific gravity of the flowing medium q, the density of fluid
DH, the pressure loss (unit: kg=m2 ) In Eq. (8), DP ¼ gDH

K 2. Friction damping
DP ¼ qV ðt ÞjV ðt Þj; (8)
2
The friction damping is induced by the friction between the fluid
where K is the pressure discharge coefficient and V is the space- and the inner wall of the moonpool, which is affected by the roughness
averaged velocity of the fluid inside the moonpool relative to the cylin- of the wall. As described in Sec. III, the inner wall of the moonpool is
der; see Sec. II B 3, Appendixes B and C. The analogy between the smooth. According to Tan et al.,29 the friction force between the
parameters and geometric configurations of the conduit and moon- hydraulic smooth wall and the fluid can be modeled by
pool is listed in Table I. For any given geometric configuration of the rffiffiffiffiffiffi
moonpool, we can get the pressure discharge coefficient by interpret- r
ing the geometric configurations into those of conduit based on Table sw ¼ q V; (9)
2
I and then referring to Table II.
As described in Sec. III, the main dimensions of the model used where sw is the shear stress on the wall, r is the angular velocity of the
in the present experiment are as follows: the cylinder has an internal fluid motion, q and  are the density and kinetic viscosities of the fluid,

27 December 2024 14:27:57

radius of a ¼ 0.2 m, external radius b ¼ 0.5 m, and draft d ¼ 0.5 m. By respectively, and V refers to the relative velocity in Eq. (8). For more
referring to Table I, we can know that the main dimensions of the con- details and discussions of the influence of wall roughness and
duit that shares the analogy of the flow field with this cylinder with Reynolds number on friction force, see Tan et al.29 and references
moonpool are the hydraulic diameter Dh ¼ 2a ¼ 1:0 m, thickness of therein.
the conduit d ¼ b  a ¼ 0:3 m; the distance between the entrance of
the conduit and the wall B ¼ d ¼ 0:5 m. Then, the inputs for Table II 3. Modified free-surface boundary condition
are d=Dh ¼ 0:3 and B=Dh ¼ 0:5, and we get the coefficient f ¼ 0:5.
Finally, we have K ¼ f ¼ 0:5 for the cylinder with the moonpool used The viscous damping force on the fluid in the moonpool is
in the experiment in Sec. III. Fd ¼ S0 DP þ sw 2pad; (10)

TABLE II. The variation of f with geometric configurations of the conduit. The data are extracted from Diagram 3-1 in Idelchik.31

Values of f

B=Dh

d=Dh 0 0.002 0.005 0.010 0.020 0.050 0.100 0.200 0.300 0.500 1

0 0.50 0.57 0.63 0.68 0.73 0.80 0.86 0.92 0.97 1.00 1.00
0.004 0.50 0.54 0.58 0.63 0.67 0.74 0.80 0.86 0.90 0.94 0.94
0.008 0.50 0.53 0.55 0.58 0.62 0.68 0.74 0.81 0.85 0.88 0.88
0.012 0.50 0.52 0.53 0.55 0.58 0.63 0.68 0.75 0.79 0.83 0.83
0.016 0.50 0.51 0.51 0.53 0.55 0.58 0.64 0.70 0.74 0.77 0.77
0.020 0.50 0.51 0.51 0.52 0.53 0.55 0.60 0.66 0.69 0.72 0.72
0.024 0.50 0.50 0.50 0.51 0.52 0.53 0.58 0.62 0.65 0.68 0.68
0.030 0.50 0.50 0.50 0.51 0.52 0.52 0.54 0.57 0.59 0.61 0.61
0.040 0.50 0.50 0.50 0.51 0.51 0.51 0.51 0.52 0.52 0.54 0.54
0.050 0.50 0.50 0.50 0.50 0.50 0.050 0.50 0.50 0.50 0.50 0.50
1 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-4

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

where S0 ¼ pa2 is the area of the inlet of the moonpool. For the fric- jVjamp . However, jVjamp itself is unknown and cannot be approxi-
tion force, only the part from the moonpool wall is considered. The mated by the results of the potential flow model. This suggests that the
details of computing V are shown in Appendixes B and C for radiation modified potential flow model may require some iterations in the
and diffraction problems, respectively. The space-averaged pressure computation. Following Tan et al.,29 the steps are given below:
discharge with Lorenz linearization is defined by DPV ¼ qV. By
Lorenz linearization, we require (i) Solve the linear potential problem formed by Eqs. (1)–(5)
by EFM. Compute jVjamp based on the results.
ðT ðT
(ii) With jVjamp obtained in (i), solve the viscous damping
Fd V ðt Þdt ¼ DPV S0 V ðt Þdt; (11) problem formed by Eqs. (1)–(4) and (12) by EFM. Compute
0 0
jVjamp again based on the results.
where T ¼ 2p=r is the period of the wave motion inside the moon- (iii) Repeat (ii) with the obtained jVjamp in the last step.
pool. Finally, we have  ¼ 12 Ks2 3p8
jVjamp þ pfw ad=S0 ; jVjamp is the
pffiffiffiffiffiffiffiffi Numerical techniques for fast convergence in Molin and
amplitude of V and fw ¼ 2r is the friction coefficient. According Legras34 section I.3 are applied for solving the unknown
to Faltinsen and Timokha,28 the modified free-surface dynamic coefficients.
boundary condition reflecting the pressure discharge is (iv) Stop the computation and output the final results when the
difference of the computed free-surface elevation at the cen-
@U DPV
þ gg ¼  ¼ V: (12) ter of the moonpool between two consecutive iterations is
@t q
lower than a threshold.

C. Solution methods
3. Numerical convergence
1. Domain decomposition and eigenfunction expansion
Convergence tests are performed during the numerical computa-
The linear potential flow problem is formed by Eqs. (1)–(5), tions. It is found that the truncation terms P, N, and Q (see
whereas the problem considering the viscous damping is formed by Appendixes B and C for the definition of the symbols) varying from
Eqs. (1)–(4) and (12). Such problems are solved using a domain- 35 to 45 are enough to get an accurate solution, with the range of the
decomposition scheme and eigenfunction matching method (EFM).

27 December 2024 14:27:57

amplitude of free-surface elevation computed by MPFM (see Secs.
As shown in Fig. 1, the whole fluid domain is divided into three sub-
IV A and IV B) lower than 0.7% at the piston-mode frequency. We
domains I, II, and III, with two adjoining boundaries T1 and T2. In
use 40 terms for P, N, and Q during the computation of MPFM. For
each subdomain, the velocity potential is written as a series of eigen-
the diffraction problem, 40 terms are used for M, P, N, and Q.
functions multiplied by unknown coefficients (see Appendixes B and
C). The eigenfunctions satisfy the governing equation (1) and corre- III. EXPERIMENTAL SETUP
sponding boundary conditions. The coefficients are determined by the
The experimental setup is illustrated in Fig. 3. The experiments
mass and pressure continuity at the boundaries; see Appendix D. The
were carried out in the multi-functional towing tank of Shanghai Jiao
procedure of finding the eigenfunctions and determining the coeffi-
cients is similar to that used by Garrett,32 Zhang et al.,8 and Han Tong University. The towing tank is 300 m long and 16 m wide, with
et al.33 the water depth being 7.5 m. The tank is equipped with a parabolic
beach and an automatically-controlled flap-type wavemaker on the
other end, as illustrated in Fig. 3. The parabolic beach is designed to
2. Scheme of iteration for searching for the solution
damp the radiated or diffracted waves. In fact, the wavemaker also has
Viscous damping depends on the amplitude of the relative veloc- the function to actively absorb the waves. The center of the cylinder is
ity between the fluid inside the moonpool and the cylinder, denoted as positioned around 100 m away from the wavemaker in the wave

FIG. 3. Sketch of the experimental setup

for a three-dimensional circular moonpool
in the multi-functional towing tank (side
view).

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-5

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

FIG. 4. Snapshots of the three-

dimensional vertical cylinder (silver) with a
moonpool rigidly connected to the
carriage.

diffraction cases, while it is positioned approximately in the middle of to the characteristic of piston-type resonance (see Molin et al.7 for the
the length of the tank in the heave radiation cases. typical modal shape of a three-dimensional circular moonpool),
Some main facilities are denoted in Fig. 4. The inner and outer although it may vary depending on the geometry and the scale of the
radii and draft of the cylinder are a ¼ 0.2 m, b ¼ 0.5 m, and d ¼ 0.5 m, moonpool foundation.
respectively. The surface of the cylinder is made of steel and is smooth. The different harmonic components of the steady-state resonant
The corners at the inlet are sharp. The cylinder undergoes small- responses inside the circular moonpool are examined by performing
amplitude oscillatory heave motions or is held fixed. The heave motion Fourier analyses on the time histories of the free-surface elevation.
was achieved by the use of an electronic motor connected to a steel Figure 7 shows the typical results at the resonance frequency, where
frame and the rail system of the carriage and recorded by the position panels (a), (c), (e), and (g) illustrate the time histories, panels (b), (d),
gauge placed on top of the vertical cylinder. The heave motion of the (f), and (h) illustrate the amplitude spectra. The mean elevation and
cylinder is achieved by the motor and the rail system. The present
focus is on the piston-mode resonance that occurs in the moonpool.

27 December 2024 14:27:57

2
The non-dimensional frequencies K ¼ rg a of the forced heave motion
and incident wave range from 0.129 to 0.455. We apply Fourier analy-
sis on samples of the time histories of the free-surface elevation to
investigate the characteristics of amplitude and frequency. The dura-
tion of each sample is the maximum value of an integer multiple of
the period of the forced heave motion or the incident wave under 80 s
and is over ten periods. Between the experimental cases, we monitored
the free-surface elevation in the moonpool and begin the next case
when its amplitude is no higher than 3 mm.
Figure 5 shows the layout of the wave probes. A total of six ultra-
sonic wave probes were adopted, named WP1–WP6, respectively. The
type of sensors assembled in wave probes is the Baumer-UNCK09T.
The wave probes have an accuracy of 0.5 mm. A position gauge, which
is placed on top of the cylinder, was adopted to measure and check the
vertical motion of the vertical cylinder. The sampling frequencies of
the wave probes and the position gauge are 100 and 1000 Hz, respec-
tively. Before the tests, all the wave probes and the position gauge were
calibrated. The uncertainty analysis of the measured free-surface eleva-
tions is discussed in Appendix A.
IV. RESULTS AND DISCUSSION
Figure 6 shows typical snapshots of the free-surface elevation
inside the moonpool at the resonance frequency. The time between
each snapshot is approximately T=4, where T is the oscillation period.
The vertical motion of the free surface can be clearly observed through
the snapshots by taking advantage of the inner wall of the cylinder as a
background.
In Secs. IV A–IV D, we only show the results based on the mea-
FIG. 5. Layout and position of wave probes for measuring the free-surface eleva-
surement of WP1 if not particularly mentioned, since we have con- tion. (a) Side view of the cylinder and (b) plan view (zoomed for the inner region).
firmed that the difference of amplitudes and phases of the free-surface WP1–WP6 are the six supersonic wave probes placed inside the moonpool. Unit:
elevations measured by different wave probes is negligible. It belongs mm.

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-6

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

FIG. 6. Snapshots of the free-surface elevation inside the moonpool from the recorded video in the heave radiation problem with n3a ¼ 10:0 mm at the resonance frequency
K ¼ 0:310. Green dashed lines are artificially drawn to show the waterline along the inner wall. (a) Zero-up crossing (t ¼ t0); (b) positive maximum (t ¼ t0 þ T=4); (c) nega-
tive maximum (t ¼ t0 þ T=2); and (d) zero-down crossing (t ¼ t0 þ 3T=4), where T is the period of the forced heave motion.

27 December 2024 14:27:57

the first and second harmonics are observed and denoted by red and phase of free-surface elevation measured in the experiment and
arrows in the figure. However, the response is generally dominated by computed by different models with respect to forcing frequencies. The
the first harmonic component. A relatively significant second har- phase of the free-surface elevation is defined relative to the forced heave
monic component can be seen in the forced motion case, as shown in motion. As can be observed, the amplitude reaches the maximum
Fig. 7(d), where the ratio of the peak of the second harmonic to the around the piston-mode resonance frequency. The resonance frequency
one of the first harmonic is around 0.026. However, for the wave dif- observed from the experimental data is around K ¼ 0:310, while the
fraction case, the maximum ratio is around 0.017, as shown in panel one predicted by using the radiation–diffraction code WAMIT is 0.312.
(h). It is known that the free-surface response amplitude of higher har- The estimated natural frequency by the single-mode approximation
monics depends on the amplitude and frequency of the incident wave model [SMA, see Eq. (2.32) in Molin et al.7] is 0.316, which is for the
or forced motion amplitude. In order to understand the source of the finite-depth water case. In addition, the one estimated by combining
mean elevation and second harmonic, we develop a time-domain sin- Eqs. (2.32) and (2.33) in Molin et al.7 is 0.298, which assumes that the
gle-degree-freedom (SDF) model to simulate the heave motion of the outer radius of the cylinder and the water depth are infinite.
water inside the moonpool. The details and results of the SDF model The phase # of the free-surface elevation is presented in Fig. 8(b).
are given in Sec. IV C. For the low-frequency region, i.e., K < 0:280, the phase is around
0:08p and increases gradually when the forcing frequency approaches
A. Heave radiation problem the piston-mode frequency. In addition, it can be observed that the
solutions by using WAMIT agree well with the experimental data in
The results using the modified potential flow model (denoted as
the non-resonant region. As can be observed, near the piston-type res-
MPFM), based on the three-dimensional pressure discharge model, are
onance frequency K ¼ 0:310, the phase is around p=2, which is simi-
compared with the measurements in the experiment and the solutions
lar to those found in the two-dimensional moonpool case (see
by a wave radiation and diffraction software called WAMIT, which is a
Faltinsen et al.11).
purely potential flow solver without accounting for viscous damping.
In order to examine the individual effect of friction and flow sep-
Details of the MPFM for the heave radiation problem are shown in
aration on the total damping, the results using the different levels of
Appendix B. Two forced heave motion amplitudes n3a ¼ 5:6 and
models are compared in Fig. 8. The MPFM contains the following
10.0 mm, corresponding to non-dimensional amplitudes e ¼ 0.028 and
three specific cases:
0.05, respectively, are adopted in the experiments with forcing frequen-
cies K ranging from 0.129 to 0.453. The frequencies in the heave radia- (i) Only consider the friction damping, without flow-separa-
tion and diffraction problems (see Sec. IV B) are chosen to be around tion-induced damping (K ¼ 0, sw 6¼ 0), denoted as MPFMf .
the piston-mode frequency to investigate the damping effect on the (ii) Only consider the flow-separation-induced damping, with-
piston-mode resonance, which is the focus of the present paper. A too out friction damping (K 6¼ 0; sw ¼ 0), denoted as MPFMv .
wide range of frequencies may result in the resonance induced by the (iii) Consider both the flow-separation-induced damping and
high-order components of the waves.35 Figure 8 shows the amplitude friction damping (K 6¼ 0; sw 6¼ 0), denoted as MPFMvf .

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-7

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

FIG. 7. Time histories of free-surface ele-

vations at the center of the moonpool
(WP1) at the resonance frequency and
corresponding amplitude spectra via
Fourier analysis are shown in the left and
right columns, respectively. Panels (a) and
(b): forced heave motion amplitude n3a ¼
5:6 mm; panels (c) and (d): n3a ¼ 10:0
mm; panels (e) and (f): incident wave
amplitude Ai ¼ 4 mm; panels (g) and (h):
Ai ¼ 8 mm. K ¼ 0:310. The red arrows
denote the mean elevation and the first

27 December 2024 14:27:57

and second harmonics.

FIG. 8. The comparison of the amplitude

and phase of free-surface elevation mea-
sured in the experiment and computed by
different numerical models for the heave
radiation problem: (a) the amplitude of the
free-surface elevation and (b) the phase
of the free-surface elevation (relative to
the forced heave motion of the cylinder).

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-8

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

We compare the results using WAMIT and MPFMf and find it

hard to distinguish them in the ranges of horizontal and vertical axes
of Fig. 8(a), and the difference of their phases is of minor importance.
Thus, the MPFMf results are not illustrated in Fig. 8 for brevity. The
viscous damping in the MPFMf is linearly proportional to V, the
averaged vertical velocity of fluid inside the moonpool relative to the
cylinder; see Eq. (9). As a result, the MPFMf is a linear model and
the non-dimensional free-surface elevation (g=n3a ) is independent of
the heave amplitude n3a , which is the same as WAMIT. The maxima
of the amplitude of the free-surface elevation of WAMIT and MPFMf
are denoted in Fig. 8(a). Comparing the results of the MPFMv and
MPFMvf , or WAMIT and MPFMf (maximum of amplitude is given
but the phase is not shown), it is found that the amplitude and phase
of the free-surface elevation change a little bit if the friction damping is
considered. It suggests that friction damping is not the dominant fac-
tor for the total viscous damping in the present experiment of heave
radiation problem.
In general, the amplitudes and phases of the free-surface eleva-
tion predicted by the present MPFMv and MPFMvf agree well with
the experiments for the two motion amplitudes, while the potential FIG. 9. The comparison of the amplitude of free-surface elevation at the center of
flow model significantly over-predicts the amplitude near the reso- the moonpool measured in the experiment (WP1) and computed by different
nance frequency such that experimental results are much lower than numerical models for the wave diffraction problem. The marks are the same as
the solutions by WAMIT. The discrepancies of phase # between the those in Fig. 8.
linear potential flow model and the experiment results are alleviated in
the MPFMv and MPFMvf by a smooth transition. This indicates that damping increases, as discussed in Tan et al.29 for a two-dimensional
flow-separation-induced damping can affect the phase and make the

27 December 2024 14:27:57

moonpool. Another reason might be that the inner wall of the moon-
variation of the phase much more smooth as the forcing frequency
pool is smooth in the present experiment. The friction damping may
approaches the piston-mode frequency. Moreover, as can be seen, the
be more significant for the real moonpool in ocean because the inner
reduction of the non-dimensional resonant wave amplitude from the
wall may not be so smooth as there might be sacrificial anodes or
linear prediction by WAMIT grows when the forced motion ampli-
marine biosorption that increases the roughness.
tude increases. This suggests that the flow-separation-induced damp-
As shown in the figure, the WAMIT over-predicts the
ing effect becomes more substantial as the excitation amplitude
responses around the piston-type resonance frequency. However,
increases.
the MPFMv and MPFMvf results show rather good agreement with
the experimental data for the two incident wave amplitudes, in par-
B. Wave diffraction problem ticular, for the case with a relatively larger incident wave amplitude
The proposed MPFM in the present study can be applied not Ai ¼ 8 mm. The deviation of the MPFMv and MPFMvf from the
only for the heave radiation problem in Sec. IV A but also for the wave experiment results is higher than that in the heave radiation prob-
diffraction problem where the cylinder is fixed in the incident wave. lem. One possible reason is that the incident wave field makes the
The numerical details on solving the wave diffraction problem are similarity of the flow fields between the diffraction problem and the
given in Appendix C. Figure 9 shows the free-surface elevation with conduit problem lower.
respect to different incident wave frequencies. Again, the experimental
results for two different incident wave amplitudes with Ai ¼ 4 and C. Different harmonic components
8 mm are compared with WAMIT and MPFM, which includes
MPFMf ; MPFMv , and MPFMvf . As the same in Sec. IV A, the differ- By applying Fourier analysis, higher harmonic components of
ence between WAMIT and MPFMf is of minor importance. Thus, the the free-surface elevation can be observed in the moonpool at the reso-
MPFMf results are not illustrated in Fig. 9, with its maximum value nance frequency, as shown in Fig. 7 and Tan et al.29 (see Fig. 4
denoted there. By comparing the MPFMv and MPFMvf results, or the therein). In the present paper, different harmonic components of the
maxima of WAMIT and MPFMf , we conclude that friction damping free-surface elevation can be observed in a wide range of frequencies,
is not the major factor of the total viscous damping. We can conclude especially around the resonance frequency. Figure 10 shows the varia-
from the results of heave radiation and wave diffraction cases in Secs. tions of gð0Þ ; gð1Þ ; gð2Þ with the frequency of the forced heave motion
IV A and IV B that in the present experiment, flow-separation-induced measured in the experiment, where the heave amplitude n3a ¼ 10:0
damping is the majority of the viscous damping, while the friction mm. Note that gð3Þ measured in the experiment is not shown in Fig.
damping is of minor importance. One of the possible reasons is that 10(d) since its amplitudes are usually lower than the accuracy of the
the sharp inlet of the moonpool in the present experiment and the wave probe. Generally, the variations of gð0Þ and gð2Þ are similar to
fluid velocity result in a strong flow separation. If the inlet of the that of gð1Þ , reaching their maxima at the resonance frequency.
moonpool is rounded, the flow separation is alleviated; thus, the ratio The possible sources of the different harmonic components
of the influence of the friction damping to the flow-separation-induced include the following:

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-9

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

FIG. 10. The amplitudes of different har-

monic components of free-surface eleva-
tion in the heave radiation problem where
n3a ¼ 10:0 mm.

27 December 2024 14:27:57

(i) The variation of the instantaneous position of the free sur- flow-separation-induced damping. The mass of the fluid in the moon-
face in the moonpool. pool MðtÞ can be computed under the following two assumptions:
(ii) The flow-separation-induced nonlinear damping
K
qjVðtÞjVðtÞ. (i) Only consider the mean mass in the moonpool,
2
MðtÞ ¼ M0 , where M0 ¼ qS0 d is the mass of fluid in the
We aim to develop a model to investigate the source of the differ- moonpool in calm water.
ent harmonic components of the free-surface elevation, for which (ii) Consider the exact mass of fluid in the moonpool,
time-domain solutions are suitable. Here, we develop a simplified MðtÞ ¼ M0 þ qSg. This reflects the effect of the varying
time-domain nonlinear model as follows. By treating the fluid in the instantaneous places of the free surface in the moonpool.
moonpool as a solid mass,7,17 the fluid in the moonpool and the plat-
form becomes a coaxial cylinder system studied by Chau and Yeung.36 To obtain the time history of free-surface elevation, Eq. (13) is
Here, we denote the inner cylinder (fluid in the moonpool) and the solved in the time domain by the fourth-order Runge–Kutta method,
outer cylinder (platform) as VC1 and VC2 , respectively. We develop a with the time step Dt ¼ T=100. The initial condition is g ¼ 0 and
dt ¼ 0, and the ramp-up is applied on the exciting force Fex ðtÞ. By
dg
single-degree-of-freedom (SDF) equation of motion for the fluid inside
the moonpool applying Fourier analysis on the time histories of the free-surface ele-
vation, we can identify the individual effects of variation of MðtÞ and
d2 g d2 g dg flow-separation-induced damping.
M ðt Þ ¼ F ex ðt Þ  l11  k11  DPmp S0  C33 g Figure 10 shows the experimental and numerical results of the
dt 2 dt 2 dt
d2 g dg K forced heave motion problem where n3a ¼ 10:0 mm, where a general
¼ Fex ðt Þ  l11 2  k11  qjV ðt ÞjV ðt ÞS0  C33 g; agreement is achieved. From Fig. 10(b), we can find that the variation
dt dt 2
of mass has minor effects on the amplitudes of the first harmonic ele-
(13)
vation. Figures 10(a) and 10(c) show that gð0Þ and gð2Þ are mainly
where MðtÞ is the mass of the fluid in the moonpool, Fex ðtÞ is the excit- induced by the variation of mass, whereas the flow-separation-induced
ing force on the
2
fluid in the moonpool, for the heave radiation problem damping has negligible effects. Moreover, we can find that the mean
Fex ðtÞ ¼ l12 ddtn23 þ k12 dn
dt ; l1j and k1j , j ¼ 1, 2 are the added mass and
3
elevation gð0Þ estimated by SDF (varying M) is higher than the experi-
damping coefficient of VC1 , respectively, induced by the motion of mental results, while the second harmonic gð2Þ by SDF (varying M) is
VCj ; S0 ¼ pa2 is the area of the inlet of moonpool [see Eq. (10)], and lower than the experimental results. We have confirmed that the same
C33 ¼ qgS0 is the hydrostatic stiffness. Here, we do not include the fric- phenomenon exists in the heave radiation problem with n3a ¼ 5:6
tional damping sw since it is of minor importance compared to the mm and the wave diffraction problems. The phenomenon may be

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-10

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

related to the assumption that simplifies the free surface in the moon- discharge coefficients from the steady flow when Kcl ranges in 0.33–
pool by a horizontal plane in the SDF model. This assumption may 3.07. In the present paper, the Kcl varies between 0.84 and 2.07, which
not reflect the nonlinear free-surface boundary condition correctly is in the range of Faltinsen and Timokha28 mentioned above. Thus, we
and thus results in the error in the energy distribution in the mean ele- also adopt the pressure discharge coefficient from the steady flow
vation gð0Þ and second harmonic gð2Þ . The discrepancies may also be through the conduit. This may underestimate the pressure discharge
contributed to the measurement error, the error of the exciting force coefficient, as Hamelin et al.39 showed that it increases with the
induced by the instantaneous position of the bottom of the platform. decreasing Kc. It is consistent with the fact that the MPFM results are
For Fig. 10(d), we only compare the theoretical results because the higher than those of the experiment in both heave radiation and dif-
experimental results are under the accuracy of the wave probe. It is fraction problems. For practical applications in ocean engineering, the
shown that the flow-separation-induced damping can induce the third values of the Keulegan–Carpenter number should be larger depending
harmonic gð3Þ and the variation of mass increases the amplitudes on the scale of the floating foundation and sea state. As a result, we
of gð3Þ . may expect the MPFM to be still valid and its error to be lower when
being applied to practical applications.
D. Implication, limitation, and future work We only studied one moonpool geometry in the present experi-
ment. As discussed above, we will conduct more validation and
We made a first attempt in analyzing the viscous damping of a
explore more about the moonpool with different geometries, the influ-
three-dimensional moonpool. It has implications on both industry and
ence of a rounded corner of the moonpool inlet, and the influence of
academy. For the industrial applications, the viscous damping coefficient
Kc in the future.
can be either used in our semi-analytical model or combined with other
potential flow software such as WAMIT to alleviate the over-estimation V. CONCLUDING REMARKS
of the amplitude of the wave in the circular moonpool. Better estima- Aiming at developing a modified potential flow model (MPFM)
tions of wave amplitudes provide reliable guidance of the design and to account for the energy losses due to both flow separation at the inlet
optimization of the floating structure at the initial stage before the dedi- of the moonpool and inner-wall friction, we investigate the piston-
cated model tests. According to the results and the corresponding analy- type resonances in a circular moonpool numerically and experimen-
ses in the present paper, we suggest using affiliates to increase the flow- tally, with the focus on the resonant wave response inside the
separation-induced damping in order to suppress the fluid motion in moonpool.

27 December 2024 14:27:57

the moonpool. For example, the fin that was experimentally investigated The MPFM is developed based on an eigenfunction-matching
by Tan et al. was used.37 For the academic research of the damping in method. A new model for flow-separation-induced viscous damping is
the moonpool with different shapes, although the coefficients in this derived by taking advantage of the similarity of the flow field of a con-
paper may not be directly used, the present research provides the follow- duit mounted in the wall. The friction on the inner wall of the moon-
ing two heuristic ways that may help: pool is also considered.
(i) Follow the method in the present paper. Search for litera- The results of the MPFM are compared with the experimental
tures studying problems that have analogy in the flow field results in terms of free-surface elevation inside the moonpool and
with the moonpool and the drag coefficients of which are phase for the heave radiation and wave diffraction problems. The
well investigated and documented. extensive comparisons suggest that the MPFM can well predict the
(ii) Conduct dedicated experiments or CFD simulations. piston-type response inside the moonpool. It was also confirmed that
Generalize the dependence of the viscous damping coeffi- the flow-separation-induced damping accounts for the majority of the
cient on the moonpool geometries. viscous damping, while the friction-induced damping is not significant
for the present inlet geometry and Kc number.
In this paper, although we focus on the case of a circular moon- Different components of the wave response in the moonpool,
pool with a sharp edge, where the flow separation point is clearly including the mean elevation and second harmonic, were observed in
defined, the derived model can be easily extended to the situation with the experiment. The conjectured sources of such components (except
a rounded corner at the inlet by extending the approach described in the first harmonic) are the variation of the instantaneous position
Tan et al.29 For the case with a rounded corner, the effect of friction- of the free surface inside the moonpool and the nonlinear quadratic
induced damping force from the inner wall side on the total damping damping induced by flow separation. Their effects are investigated by
may be enhanced depending on the corner radius, which should be time-domain simulations based on a single-degree-of-freedom model
investigated in the future. of the fluid inside the moonpool. The results show that the variation of
As mentioned by many literatures, the pressure discharge coeffi- the mass of the fluid inside the moonpool generates the majority of
cient may vary with the Keulegan–Carpenter number Kc (see, e.g., the different harmonic components, whereas the flow-separation-
Faltinsen and Timokha,28 Tan et al.,29 Molin38). In the present paper, induced damping is of minor importance.
we define the local Keulegan–Carpenter number as It should be noted that the Kc values in the practical application
2pg may be larger and its effect should be studied further to improve Eq.
Kcl ¼ ; (14) (8) with the dependence of the pressure discharge coefficient K on Kc,
L=2
in particular, for the case with a rounded corner near the moonpool
where g is the amplitude of the free-surface elevation in the moonpool, inlet.
L is the width of the moonpool, for the circular moonpool in the pre- Despite the experimental and numerical analyses of the viscous
sent paper, L ¼ 2a. Faltinsen and Timokha28 adopt the pressure energy losses and different harmonic components in the moonpool,

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-11

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

we believe that the method in the present paper can be extended to TABLE III. The bias limits for the sources of uncertainty for

g in experiments of the
cases with a rounded corner at the inlet following the approach in Tan heave radiation problem with n3a ¼ 10:0 mm, K ¼ 0.235 and 0.310.
et al.29 As the same with many other literatures (see, e.g., Faltinsen and
Timokha28 and Tan et al.29), we also emphasize the importance of K Source of uncertainty Bj hj Bj hj
investigating the dependence of the pressure discharge coefficient on 0.235 g (m) 5:00  104 101.05 0.05
the Keulegan–Carpenter number when applying this method in prac-
n3a (m) 2:00  105 –121.47 –0.002
tical applications.
0.310 g (m) 5:00  104 101.06 0.05
n3a (m) 2:00  105 –679.84 –0.01
ACKNOWLEDGMENTS
The authors acknowledge the technical assistance from H. Xu
during the experiments in the towing tank in SJTU. The present work We denote the non-dimensional free-surface elevation at the center
is sponsored by the Oceanic Interdisciplinary Program of Shanghai Jiao of the moonpool by
g ¼ g=n3a .
Tong University (Project No. SL2021PT205). The corresponding The bias limit of a result r computed by data reduction equa-
author (X. Zhang) gratefully acknowledges the financial support from tion r ¼ rðX1 ; X2 ; …; XN Þ is evaluated by
the National Natural Science Foundation of China under Grant No.
52171269. X
N
B¼ h2j B2j ; (A1)
j¼1
AUTHOR DECLARATIONS
where N is the number of the sources of the uncertainty for r. Bj is
Conflict of Interest
the bias limit for the jth source of uncertainty. hj ¼ @r=@Xj is the
The authors have no conflicts to disclose. corresponding sensitivity coefficient. The bias limits for the sources
of uncertainty for the non-dimensional free-surface elevations are
Author Contributions presented in Table III.
Jian Han: conceptualization (equal); formal analysis (equal); investiga- The precision limit for multiple tests is evaluated by
tion (equal); methodology (equal); validation (equal); writing— Kcf S
g

27 December 2024 14:27:57

original draft (equal); and writing—review and editing (equal). Xinshu PðMÞ ¼ pffiffiffiffiffi ; (A2)
Zhang: conceptualization (lead); formal analysis (equal); funding M
acquisition (equal); investigation (equal); methodology (equal); super- where S
g is the standard deviation evaluated based on multiple
vision (equal); validation (equal); writing—original draft (equal); and runs. M is the number of repeated tests. Kcf is the coverage factor,
writing—review and editing (equal). Haiyang Huang: data curation taken from Student’s t-distribution for a 95% confidence level. For
(equal); formal analysis (equal); investigation (equal); methodology M  10, Kcf is taken to be 2.
(equal); validation (equal); and writing—original draft (equal). The uncertainty for the non-dimensional free-surface eleva-
tions is evaluated by
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
DATA AVAILABILITY
u
g ¼ B2 þ PðMÞ2 : (A3)
The data that support the findings of this study are available
from the corresponding author upon reasonable request. The results of the uncertainty analysis for the non-dimensional
free-surface elevations are listed in Table IV, where ~ g is the mean
value of the non-dimensional free-surface elevations based on the
APPENDIX A: UNCERTAINTY ANALYSES repeated tests.
AND REPEATABILITY As shown in Table IV, the uncertainty level is rather satisfac-
The test repeatability was checked at the beginning and tory. It also suggests that the precision limits in cases of both
throughout the testing program. In order to analyze the uncertainty frequencies contribute to the main uncertainty for the non-
during the experiment, we performed ten tests for particular cases. dimensional free-surface elevations. Case A corresponding to the
The present uncertainty analyses are performed by following the lower non-dimensional free-surface elevation shows a higher uncer-
ITTC guideline.40 tainty level. We also confirm that durations of time traces have little
In the present study, it is not practical to evaluate the uncer- effect on the uncertainty levels.
tainty of the non-dimensional free-surface elevations for each tested
case, which requires a large number of tests for various frequencies
TABLE IV. Evaluation of the uncertainty for

g for experiments of the heave radiation
and forced heave motion or incident wave amplitudes. Therefore, problem with n3a ¼ 10:0 mm ; K ¼ 0.234 and 0.310.
the uncertainty of the free-surface elevations is evaluated for the
cases of heave radiation problem where the heave amplitude is Cases Bias limit Precision limit u
g ~g u
g =~g
n3a ¼ 10:0 mm, with two forcing frequencies being K ¼ 0:234 (case
A) and K ¼ 0:310 (case B), respectively. The latter corresponds to A 0.003 0.031 0.031 1.20 2.58%
the largest amplitude of free-surface elevation inside the moonpool. B 0.003 0.036 0.036 6.73 0.53%
Uncertainty tests are repeated ten times for both case A and case B.

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-12

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

APPENDIX B: MPFM FOR THE HEAVE RADIATION The space-averaged velocity of the fluid inside the moonpool rela-
PROBLEM tive to the cylinder is
We define the heave displacement of the platform as V ¼ irAi veirt ; (C4)
  Ð a @w0
n3 ¼ < n3a eirt (B1) where v ¼ S10 2p 0 @z z¼0 rdr.
and denote the velocity potential U ¼ <½irn3a uðr; zÞeirt , where The solution in subdomain I is
u is the spatial potential. X
1
iv
The space-averaged velocity of the fluid inside the moonpool uI ¼ bn in wInðDÞ cos ðnhÞ  (C5)
relative to the cylinder is n¼0
r
ð ð !
1 2p a @U dn3 with
V¼  rdrdh ¼ irn3a veirt ; (B2)
S0 0 0 @z z¼0 dt X
1
wIðDÞ
n ¼ Fnk Cnk ðrÞZkI ðzÞ: (C6)
where k¼0
ð 2p ð a
1   The solution in subdomain II is
v¼ uz jz¼0  1 rdrdh: (B3)
S0 0 0 X
1
 
wIIn ¼ Dnk Rnk ðr Þ þ Dnk Rnk ðr Þ ZkII : (C7)
The solutions of spatial potentials at each subdomain are k¼0
X
1
iv
uI ¼ Ak Ck ðrÞZkI ðzÞ  ; (B4) The spatial potential of the incident wave is
r
g coshm0 ðz þ hÞ X
1
k¼0
X
1
  II u0 ¼ b in Jn ðm0 r Þcos ðnhÞ: (C8)
u ¼
II
Cn Rk ðr Þ þ Ck Rn ðrÞZk ðzÞ þ uIIp ; (B5) r 2 coshm0 h n¼0 n
k¼0
X
1 The solution in subdomain III is
uIII ¼ Bk !k ðrÞZkI ðzÞ:

27 December 2024 14:27:57

(B6) X1
g coshm0 ðz þ hÞ
n ¼
wIII Jn ðm0 r Þ þ Enk !nk ZkI :
k¼0 (C9)
r2 coshm0 h
uIIp is the particular solution and is written as k¼0

Fnk, Dnk, Dnk , and Enk are the unknown coefficients, which will
1 r2
uIIp ¼ ðz þ hÞ2  ; (B7) be determined by applying the matching conditions described in
2ðh  dÞ 2 Appendix D; Cnk ðrÞ; Rnk ðrÞ; Rnk ðrÞ, and !nk ðrÞ are the eigenfunc-
where Ak , Ck , Ck , and Bk are the unknown coefficients, which will tions in the radial direction; ZkI and ZkII are the eigenfunctions in the
be determined by applying the matching conditions described in vertical direction. For the details of the functions, please refer to
Appendix D; Ck ðrÞ; Rn ðrÞ; Rn ðrÞ, and !k ðrÞ are the eigenfunctions Mavrakos.12 The truncation of the infinite series is the same as
in the radial direction; ZkI and ZkII are the eigenfunctions in the ver- those described in Appendix B. In the matching procedure of the
tical direction. For the details of the functions, please refer to Chau numerical model, the infinite series of uI ; uII , and uIII at truncated
and Yeung36 and Mavrakos.13 In the matching procedure of the at the starting M, N, and Q terms, i.e., n 2 ½0; M  1, for Eq. (C3).
numerical model, the infinite series of uI ; uII , and uIII at truncated The infinite series of wI, wII, and wIII are truncated at the starting P,
at the starting P, N, and Q terms, i.e., k 2 ½0; P  1; ½0; N  1, and N, and Q terms, i.e., k 2 ½0; P  1; ½0; N  1, and ½0; Q  1 for
½0; Q  1 for Eqs. (B4), (B5), and (B6), respectively. Eqs. (C6), (C7), and (C9), respectively.

APPENDIX C: MPFM FOR WAVE DIFFRACTION APPENDIX D: MATCHING CONDITIONS

PROBLEM
The matching conditions on the common boundaries T1 and
The incident wave is denoted as T2 between the three subdomains are written as
 
gi ¼ < Ai eiðkxrtÞ ; (C1) uI ¼ uII on T1 ; (D1a)
where Ai is the amplitude of the incident wave. The velocity poten- 8
tial in the fluid domain is @uI < 0 II on r ¼ a; d < z < 0;
¼ @u (D1b)
  @r : on T1 ;
U ¼ < irAi uðr; h; z Þeirt ; (C2) @r
uIII ¼ uII on T2 ; (D1c)
where u is the spatial potential. It is further written as 8
X
1 @uIII < 0 II on r ¼ b; d < z < 0;
u¼ bn in wn cos ðnhÞ: (C3) ¼ @u (D1d)
@r : on T2 :
n¼0 @r

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-13

Published under an exclusive license by AIP Publishing
 Physics of Fluids ARTICLE pubs.aip.org/aip/pof

By substituting the solutions of each subdomain into the matching 20

A. G. Fredriksen, T. Kristiansen, and O. M. Faltinsen, “Wave-induced response
conditions and taking advantage of the orthogonality of the eigen- of a floating two-dimensional body with a moonpool,” Philos. Trans. R. Soc. A
373, 20140109 (2015).
functions in the vertical direction, the unknown coefficients can be 21
S.-C. Jiang, W. Bai, and G. Tang, “Numerical investigation of piston-modal
obtained. Furthermore, the free-surface elevation inside the moon- wave resonance in the narrow gap formed by a box in front of a wall,” Phys.
pool and the local fluid velocity can be computed. Details of the Fluids 31(5), 052105 (2019).
procedure can be referred to in Zhang and Bandky,41 Zhang and 22
T. Kristiansen and O. M. Faltinsen, “Application of a vortex tracking method
Yeung,42 and Han et al.33 to the piston-like behaviour in a semi-entrained vertical gap,” Appl. Ocean Res.
30, 1–16 (2008).
23
T. Kristiansen and O. M. Faltinsen, “Gap resonance analyzed by a new
REFERENCES domain-decomposition method combining potential and viscous flow draft,”
1
B. Molin, I. de Vries, and A. Cinello, “Hydrodynamic analysis of the piston mode Appl. Ocean Res. 34, 198–208 (2012).
resonance inside a large flng turret,” in Proceedings of the 29st International 24
J. K. Heo, J. C. Park, W. C. Kim, and M. H. Kim, “Influences of vorticity to ver-
Workshop on Water Waves and Floating Bodies, Osaka, Japan (2014). tical motion of two-dimensional moonpool under forced heave motion,” Math.
2
H. Gupta, R. Blevins, and H. Banon, “Effect of moonpool hydrodynamics on Probl. Eng. 2014, 1–13.
spar heave,” in Proceedings of the 27th International Conference on Ocean, 25
B. Molin, F. Remy, O. Kimmoun, and Y. Stassen, “Experimental study of the
Offshore and Arctic Engineering (OMAE), Estoril, Portugal (2008). wave propagation and decay in a channel through a rigid ice-sheet,” Appl.
3
L. Guignier, A. Courbois, R. Mariani, and T. Choisnet, “Multibody modelling Ocean Res. 24(5), 247–260 (2002).
of floating offshore wind turbine foundation for global loads analysis,” in 26
B. Molin and F. Remy, “Experimental and numerical study of the gap resonan-
Proceedings of the 26th International Ocean and Polar Engineering Conference, ces in-between two rectangular barges,” in Proceedings of the 13th
Rhodes, Greece (2016). International Congress International Maritime Association of Mediterranean
4
C. Wang, S. Zheng, and Y. Zhang, “A heaving system with two separated oscil- (IMAM), Istambul, Turkey (2009).
lating water column units for wave energy conversion,” Phys. Fluids 34(4), 27
X. B. Chen, H. X. Liu, and W. Y. Duan, “Semi-analytical solutions to wave dif-
047103 (2022). fraction of cylindrical structures with a moonpool with a restricted entrance,”
5
T. A. Johansen, T. I. Fossen, S. I. Sagatun, and F. G. Nielsen, “Wave synchro- J. Eng. Math. 90, 51–61 (2015).
nizing crane control during water entry in offshore moonpool operations- 28
O. M. Faltinsen and A. N. Timokha, “On damping of two-dimensional piston-
experimental results,” IEEE J. Oceanic Eng. 28(4), 720–728 (2003). mode sloshing in a rectangular moonpool under forced heave motions,”
6
B. Molin, “On the piston sloshing modes in moonpools,” J. Fluid Mech. 430, J. Fluid Mech. 772, R1 (2015).
27–50 (2001). 29
L. Tan, L. Lu, G.-Q. Tang, L. Cheng, and X.-B. Chen, “A viscous damping
7
B. Molin, X. Zhang, H. Huang, and F. Remy, “On natural modes in moonpools model for piston mode resonance,” J. Fluid Mech. 871, 510–533 (2019).

27 December 2024 14:27:57

30
and gaps in finite depth,” J. Fluid Mech. 840, 530–554 (2018). R. D. Blevins, Applied Fluid Dynamics (Krieger, 1992).
31
8
X. Zhang, H. Huang, and X. Song, “On natural frequencies and modal shapes I. Idelchik, Handbook of Hydraulic Resistance (Begell House, 1996).
in two-dimensional asymmetric and symmetric moonpools in finite water 32
C. J. R. Garrett, “Wave forces on a circular dock,” J. Fluid Mech. 46, 129–139
depth,” Appl. Ocean Res. 82, 117–129 (2019). (1971).
9
B. Chu and X. Zhang, “On the natural frequencies and modal shapes in two- 33
J. Han, X. Zhang, and R. W. Yeung, “Hydrodynamic behavior of a circular
dimensional moonpools with recesses in finite water depth,” Appl. Ocean Res. floating solar pond with an entrapped two-layer fluid,” Phys. Fluids 34(1),
115, 102787 (2021). 012114 (2022).
10
X. Zhang and Z. Li, “Natural frequencies and modal shapes of three- 34
B. Molin and J. L. Legras, “Hydrodynamic modeling of the Roseau tower stabi-
dimensional moonpool with recess in infinite-depth and finite-depth waters,” lizer,” in Proceedings of the 9th OMAE Conference (ASME, 1990), pp. 329–336.
Appl. Ocean Res. 118, 102921 (2022). 35
S.-C. Jiang, W. Bai, and B. Yan, “Higher-order harmonic induced wave reso-
11
O. M. Faltinsen, O. F. Rognebakke, and A. N. Timokha, “Two-dimensional res- nance for two side-by-side boxes in close proximity,” Phys. Fluids 33(10),
onant piston-like sloshing in a moonpool,” J. Fluid Mech. 575, 359–397 (2007). 102113 (2021).
12
S. A. Mavrakos, “Wave loads on a stationary floating bottomless cylindrical 36
F. P. Chau and R. W. Yeung, “Inertia, damping, and wave excitation of heaving
body with finite wall thickness,” Appl. Ocean Res. 7(4), 213–224 (1985). coaxial cylinders,” in Proceedings of the 31st ASME International Conference on
13
S. A. Mavrakos, “Hydrodynamic coefficients for a thick-walled bottomless cylin- Ocean, Offshore and Arctic Engineering (ASME, 2012), pp. 803–813.
drical body floating in water of finite depth,” Ocean Eng. 15(3), 213–229 (1988). 37
L. Tan, L. Cheng, and T. Ikoma, “Damping of piston mode resonance between
14
X. Feng and W. Bai, “Wave resonances in a narrow gap between two barges using two fixed boxes,” Phys. Fluids 33(6), 062117 (2021).
fully nonlinear numerical simulation,” Appl. Ocean Res. 50, 119–129 (2015). 38
B. Molin, “Hydrodynamic modeling of perforated structures,” Appl. Ocean
15
Z. Song, L. Lu, L. Cheng, Y. Liu, G. Tang, and X. Lou, “Fully nonlinear numeri- Res. 33(1), 1–11 (2011).
cal investigations on the dynamics of fluid resonance between multiple bodies 39
J. Hamelin, J. Love, M. Tait, and J. Wilson, “Tuned liquid dampers with a
in close proximity,” Phys. Fluids 34(12), 122106 (2022). Keulegan–Carpenter number-dependent screen drag coefficient,” J. Fluids
16
E. Hammargren and J. K. Tornblom, “Effect of the moonpool on the total resis- Struct. 43, 271–286 (2013).
tance of a drillship,” Master’s thesis (Chalmers University of Technology, 40
ITTC, “Testing and extrapolation methods: Resistance uncertainty analysis,
Goteborg, Sweden, 2012). example for resistance test (7.2-02-02-02),” ITTC-Recommended Procedures
17
B. Molin, “On natural modes in moonpools with recesses,” Appl. Ocean Res. and Guidelines (ITTC, 2002), pp. 1–18.
67, 1–8 (2017). 41
X. Zhang and P. Bandyk, “On two-dimensional moonpool resonance for twin
18
J. N. Newman, “Resonant response of a moonpool with a recess,” Appl. Ocean bodies in a two-layer fluid,” Appl. Ocean Res. 40, 1–13 (2013).
Res. 76, 98–109 (2018). 42
X. Zhang and R. W. Yeung, “On hydrodynamic behavior of a cylindrical
19
X. Xu, X. Zhang, B. Chu, and H. Huang, “On natural frequencies of three- moonpool with an entrapped two-layer fluid,” in Proceedings of the 31st
dimensional moonpool of vessels in the fixed and free-floating conditions,” International Workshop on Water Waves and Floating Bodies, Plymouth, MI
Ocean Eng. 195, 106656 (2020). (2016).

Phys. Fluids 35, 082106 (2023); doi: 10.1063/5.0160461 35, 082106-14

Published under an exclusive license by AIP Publishing